[Insert Fig. 6 here]
3.2 Validating DRM with the lysimetry experiment
Peak transpiration fluxes were 5.5, 6 and 7 kg h-1, for Trees #1, #2 and #3, respectively, from Feb to May 2016. The peak temperature rise of Probe #2 was around 2 K at midday and 0.6 K at night. The peak sap velocity calculated by the DRM was > 60 cm h-1 for Tree #1, 80 cm h-1 for Tree #2 and 100 cm h-1 for Tree #3.
The DRM reproduced diel patterns of sap flow measured by lysimetry under a wide range of flow conditions, from negligible flows at night to very large and fluctuating flows (see Fig. 7 for the sap flow, Fig.8 for the sap velocity of Tree #3 and Fig. S5 for all the three trees). The HRM, by contrast, was unable to capture sap velocity > 25 cm h-1 (> 2 kg h-1, Fig. 7 ). While better than the HRM in capturing large sap velocities (Fig. 7 ,8 ), the CHPM also failed when velocity was either very high or very low (see Fig.8 ). The Tmax method failed to capture sap velocities < 20 cm h-1. To match peak velocity measured by lysimetry (see Table S3 ), scaling factors for Tmax had to be increased with time. For example, scaling factors increased 1.3-fold from Feb to Apr, and 1.6-fold from Feb to May. The sensitivity of the Tmax method to noise and probe alignment requires constant adjustment of scaling factors as trees grow.